What are “ultraprocessed foods” and are they a health risk? One would think there is reason for concern, given all the hyped headlines, but as is often the case the data is somewhat complicated.
The complexity begins with the fact that there is no agreed-upon operational definition of “ultraprocessed food” (UPF). Processed food generally refers to food products that are already prepared and packaged, as opposed to raw ingredients that need to be prepared and cooked. This could include a can of soup or a baked good. There is no clear dividing line between processed and ultraprocessed, and no specific features that are considered key.
One of the common features of processed foods is that they have additives which serve a variety of purposes. These include preservatives to prolong shelf life and enhance food safety, stabilizers, emulsifiers, anti-oxidants, flavoring, and coloring agents. Each ingredient, of course, is studied and approved separately by the FDA or similar agencies in other countries.
Processed foods also tend to be higher in fat, sugar, and salt, and lower in fiber and micronutrients than foods prepared from raw ingredients. Diets with such a profile are already known to carry health risks.
There have now been numerous studies, all observational, looking to see if there is any correlation between the consumption of processed or ultraprocessed foods and adverse health outcomes, including all-cause mortality. Most prominently is data from the French online health survey NutriNet-Santé which has spawned multiple studies. One study published in JAMA found:
A total of 602 deaths (1.4%) occurred during follow-up. After adjustment for a range of confounding factors, an increase in the proportion of ultraprocessed foods consumed was associated with a higher risk of all-cause mortality (HR per 10% increment, 1.14; 95% CI, 1.04-1.27; P = .008).
So it looks like consumption of upf was association with a 10% increase in all cause mortality. But many experts have found this data unconvincing, mostly for the usual reasons that limit the interpretation of observational studies. In this data set consumption of upf was associated with lower income, lower educational level, living alone, higher BMI and lower physical activity. These are all features that increase the risk of all-cause mortality. When you account for preexisting conditions the statistical significance of the effect is reduced, and if you eliminate those who die in the first year of the study, it goes away.
One way to look at this data is that, when we study each individual ingredient that may be present in processed food they appear to be safe. When you look at processed food consumption itself (rather than specific ingredients) you observationally see a correlation with negative health outcomes, but this data is massively confounded by a host of lifestyle factors.
Overall it seems highly likely that increased consumption of processed food is simply a marker for people who need the convenience of preprepared food because they live alone and are low SES (socioeconomic status). People in this category tend to be more overweight and exercise less, in addition to a host of other health factors.
But perhaps the biggest weakness of this data is the way it chooses to define processed and ultraprocessed foods, by the number of ingredients they have that are typically used in processed food. The nutritional quality of the food is generally not considered, nor are the ingredients themselves or their purpose.
The result is a focus on the wrong thing – where and how food is prepared rather than its nutritional content. Processed foods may also be high in fat, sugar, and salt, but so can non-processed or minimally processed foods. Also, telling people not to eat cheap, tasty, convenient food seems unlikely to be a successful public health measure. The people who currently rely on cheap convenient food likely don’t have many other options.
A better approach, one that is generally already used, is to investigate individual ingredients for their safety. Second, focus on public education regarding how to have a healthful diet. Eat a varied diet with plenty of foods that are high in nutrients, such as fruits and vegetables. Keep an eye on your overall salt and sugar intake. Keep it simple and focus on the data that is highly reliable.
There can also be a positive impact from focusing on the manufacturing end – make food labeling informative and easy to read and understand. Don’t allow deceptive or manipulative labeling. Other public health measures that may be effective include addressing so-called food deserts, where fresh foods are simply not locally available.
Demonizing processed foods based upon weak observational data that focuses on the wrong thing, however, is unlikely to be helpful.